Create Graphs/Analyses without Performing Rarefaction

[pkirti33]

Hello,
I acknowledge that I have submitted several posts to this GitHub Discussions in the past few weeks, and I always appreciate your prompt and helpful replies! This package and your replies have been incredibly helpful for my analysis.

Feedback
I am analyzing presence/absence microbiome data, so rarefaction is not appropriate for my project. However, I noticed that rarefaction automatically occurs when I run any analyses or create graphs. Is there a way to turn off the automatic rarefaction? If not, adding that option would be incredibly useful.

[cafferychen777]

Hello @pkirti33,

Thank you for your continued engagement with MicrobiomeStat and for bringing this important point to our attention. We greatly appreciate your feedback as it helps us improve the package for all users.

I'm pleased to inform you that we have implemented changes based on your suggestion. We have modified the package so that automatic rarefaction no longer occurs by default for most functions. Here are the key points of the update:

1. Automatic Rarefaction Removed: We have removed automatic rarefaction from all analysis and graphing functions, with the exception of the report functions.

2. Report Functions: The report functions will still perform rarefaction by default. This is to ensure consistency across different analyses within a single report.

3. Flexibility: This change provides more flexibility for users working with various types of microbiome data, including presence/absence data like yours.

We believe this update will better serve your needs and those of other researchers working with non-count data. You should be able to run your analyses and create graphs without any automatic rarefaction interfering with your presence/absence data.

If you encounter any issues or have any questions about using the updated functions, please don't hesitate to reach out. We're here to help ensure your analysis goes smoothly.

Thank you again for your valuable feedback. It's through user input like yours that we can continue to improve MicrobiomeStat and make it more versatile for different types of microbiome studies.

Best regards,
Chen YANG

[pkirti33]

Hello,
I'm back to using the package and see that the change has been made. Thank you so much for the help!

I noticed that running generate_taxa_barplot_single() performs TSS by default. Is it possible for the function to perform no rarefaction and no normalization?

[cafferychen777]

Dear pkirti33,

Thank you for your continued use of MicrobiomeStat and for your feedback. I'm glad to hear that the changes we've made have been helpful for your analysis.

Regarding your question about generate_taxa_barplot_single() and TSS (Total Sum Scaling) normalization, you're correct that it's performed by default. While it's possible to modify the function to skip normalization, I believe retaining TSS is generally beneficial for several reasons:

1. Comparability: TSS ensures that all samples are scaled to the same total, making it easier to compare relative abundances across samples or groups.

2. Visual consistency: Without normalization, the height of bars for each sample or group in the plot could vary significantly, potentially making the visualization less intuitive to interpret.

3. Addressing compositional nature: Microbiome data is inherently compositional, and TSS helps to account for this by focusing on relative rather than absolute abundances.

However, I understand that different analyses may have specific requirements. If you have a particular reason for avoiding TSS in your barplots, please let me know, and we can discuss potential alternatives or modifications to the function.

Is there a specific concern you have about using TSS for your presence/absence data? I'd be happy to explore this further with you to ensure MicrobiomeStat meets your analytical needs.

Best regards,
Chen

[pkirti33]

Hi,
Thank you for your response and continued help! My main concern with TSS is that it scales the values in each column (so that they sum to a consistent total). However, since my data is binary (representing presence/absence), scaling this data to values other than 0 or 1 would lead to data that doesn't really make sense.